The present invention relates to improving the bioactivity of a diverse range of peptides and proteins in their various forms (collectively "polypeptides") by employing conformation-constraining residues to flank sites of the polypeptide that are involved in protein-protein interactions.
Protein-protein interactions are crucial to almost every physiological and pharmacological process. These interactions often are characterized by very high affinity, with dissociation constants in the low nanomolar to subpicomolar range. Such strong affinity between proteins is possible when a high level of specificity allows subtle discrimination among closely related structures. The interaction sites of several protein pairs have been identified by strategies such as chemical modification of specific amino acid residues, site-directed mutagenesis, peptide synthesis, X-ray diffraction studies and theoretical approaches.
Certain general structural features have emerged from these studies. For example, some interactions involve more than one interaction site. The phrase "interaction site" is used in this description to denote a site comprised of amino acid residues which is involved in the interaction between two proteins. The high affinities at these interaction sites are attributed to several factors, including shape complementarity, electrostatic and hydrogen bond links, and burial of hydrophobic groups. A protein-protein interaction may involve one or more of these factors at each interaction site.
The amino acids of an interaction site usually constitute a small proportion of the total amino acids present in the polypeptide. Typically, the number of amino acid residues in a single interaction site ranges from three to six. These residues often are connected by the peptide bonds of adjacent residues in a continuous interaction site. Alternatively, the amino acid residues involved in the interaction are not linked directly by peptide bonds, but rather are brought together by the three-dimensional folding of the protein and are known as "discontinuous" sites. Due to this extensive variability, it has been difficult to identify the amino acids of interaction sites.
The chemical nature of the side chains of the amino acid residues contributes significantly to the interaction, although main chain atoms also can be involved. Positively charged residues (such as lysine, arginine and histidine) can associate through salt bridge links with negatively charged residues (such as aspartic acid and glutamic acid). Additionally, the side chains of leucine, isoleucine, methionine, valine, phenylalanine, tyrosine, tryptophan and proline are often involved in hydrophobic interactions. Precise alignment of atoms between the interaction sites of one protein and its partner also allow multiple Van der Waals interactions and thus increase the likelihood of strong binding between the two interaction partners.
Bernstein et al., Nature 340: 482 (1989), proposed a role for methionine in protein-protein interactions, whereby clusters of methionine residues in the 54,000 MW signal recognition particle play a key role in the recognition of signal peptides. Because of the unique flexibility of their side chains, methionine residues located on one face of the surface of the amphiphilic helix provide a malleable, nonpolar surface. It was postulated that in the binding process, this surface can adapt itself to peptide partners of various dimensions and thus conform to the structure of the signal peptide. A similar mechanism was suggested for the ability of calmodulin to interact with various protein partners. The binding site of calmodulin contains eight exposed methionine residues. Such flexibility of the side chain of methionine might be attributable to the presence of the sulfur atom.
A large number of proteins are synthesized as inactive precursors and activated in vivo only where and when they are needed. Accordingly, their activity is strictly regulated so as to contribute to the overall control of physiological processes. Some of these proteins are activated by the action of specific proteinases, and the interaction between the cleavage site and the proteinases should be deemed highly specific. Therefore, the regions around these activation sites form another group of protein-protein interaction sites.
As described above, the diverse properties of the various amino acids affect the characteristics of the interaction site, as well as the polypeptide as a whole. One amino acid residue that has wholly unique structural characteristics is proline.
Proline is the only common imino acid found in proteins. The side chain of proline is bonded to the tertiary nitrogen in a cyclic pyrrolidine ring. This ring inhibits free rotation about the C.sub..alpha. --N bond and thus restricts the range of allowable conformations of the polypeptide backbone. The pyrrolidine ring also constrains the conformation of the adjacent residues. The imino nitrogen of the proline residue lacks a proton that is required for hydrogen bond formation in both the .alpha.-helical and .beta.-pleated sheet conformations. Accordingly, proline is often called a "helix breaker." Additionally, the carbonyl oxygen atom of the amino acid residue immediately preceding proline in the polypeptide is more electronegative than carbonyl oxygen atoms preceding other amino acid residues. As a result, this carbonyl group has an enhanced tendency to accept and form strong hydrogen bonds.
Proline also differs from other amino acid residues in terms of permissible bond configuration. The partial double bond character of peptide bonds prevents free rotation and can result in either cis or trans configurations around the peptide bond. In the cis configuration, the C.sub..alpha. atoms of adjacent amino acid residues are closer than in the trans configuration. This "closeness" often causes steric hindrance between the side chains on the two C.sub..alpha. atoms. Accordingly, almost all peptide bonds are in the trans configuration, so that the C.sub..alpha. atoms of adjacent amino acid residues are separated by the greatest distance possible. In contrast to most amino acids, proline residues can more readily assume cis configurations because the amide nitrogen is part of a ring. Of course, the ring still imposes conformational constraints by inhibiting free rotation around the .alpha. carbons of adjoining residues.
Previous studies have implicated praline residues at some interaction sites of certain classes of molecules. Praline has been thought to be required in the interaction site geometry of a class of proteins known as the serine proteinase inhibitors. This proposal was later retracted because praline was not universally present near interaction sites. See Laskowski and Kato, Ann. Rev. Biochem. 49: 593-626 (1980). A proline-directed arginyl cleavage at monobasic processing sites has also been proposed. See Schwartz, FEBS Letts. 200: 1-10 (1986). About a third of the monobasic processing sites contain a praline residue either just before or just after the basic residue. A praline residue was also found to be important in the processing of the signal peptide of human lysozyme. Several proline-directed kinases which phosphorylate their substrates at the residues that are immediately followed by proline residues have been purified from various sources. In some cases, a proline residue two or three residues before the phosphorylation site also appears to have importance. But these phosphorylation sites include only a few examples. Hence, it was assumed heretofore that proline was involved in only a small number of specific cases.
Proline is known to be a helix breaker because it has a secondary amine group, which cannot form hydrogen bonds with neighboring CO groups as other amino acids do. Still, proline is found in some of the surface helices of soluble proteins. The "kink" induced by the proline may help in helical packing by wrapping the helix around a protein core. Recently, the importance of proline residues in transmembrane helices has been noted. Interestingly, the putative transmembrane helices of ion channel peptides have a proline residue within their sequence. These proline residues tend to be conserved among homologous proteins, while similar transmembrane helices of non-transport proteins seem to be devoid of proline residues. The convex side of a proline-containing helix is packed against neighboring transmembrane helices. The unique geometry of the proline frees a non-hydrogen bonded carbonyl oxygen in the helix backbone for binding to a cation.
Because an interaction site of a polypeptide is so difficult to identify, the interaction sites of most proteins have remained unknown. In one of its aspects, the present invention takes advantage of proline positioning to identify interaction sites of proteins. The unique properties of proline have a stabilizing function which, according to another aspect of the present invention, can be used to engineer novel polypeptides based on biologically-active polypeptides. These biologically-active polypeptides possess a functional activity and include naturally-occurring polypeptides or polypeptides derived therefrom. These novel polypeptides can have improved activities, stabilities or other properties of interest.